No abstract

Landscape degradation is a major threat to global biodiversity that is being further exacerbated by climate change. Halting or reversing biodiversity decline using seed-based restoration requires tons of seed, most of which is sourced from wild populations. However, in regions where restoration is most urgent, wild seed sources are often fragmented, declining and producing seed with low genetic diversity. Seed production areas (SPAs) can help to reduce the burden of collecting native seed from remnant vegetation, improve genetic diversity in managed seed crops and contribute to species conservation. Banksia marginata (Proteaceae) is a key restoration species in south-eastern Australia but is highly fragmented and declining across much of its range. We evaluated genetic diversity, population genetic structure and relatedness in two B. marginata SPAs and the wild populations from which the SPA germplasm was sourced. We found high levels of relatedness within most remnants and that the population genetic structure was best described by three groups of trees. We suggest that SPAs are likely to be important to meet future native seed demand but that best practice protocols are required to assist land managers design and manage these resources including genetic analyses to guide the selection of germplasm.

Monitoring vegetation restoration is challenging because monitoring is costly, requires long‐term funding, and involves monitoring multiple vegetation variables that are often not linked back to learning about progress toward objectives. There is a clear need for the development of targeted monitoring programs that focus on a reduced set of variables that are tied to specific restoration objectives. In this paper, we present a method to progress the development of a targeted monitoring program, using a pre‐existing state‐and‐transition model. We (1) use field data to validate an expert‐derived classification of woodland vegetation states; (2) use these data to identify which variable(s) help differentiate woodland states; and (3) identify the target threshold (for the variable) that signifies if the desired transition has been achieved. The measured vegetation variables from each site in this study were good predictors of the different states. We show that by measuring only a few of these variables, it is possible to assign the vegetation state for a collection of sites, and monitor if and when a transition to another state has occurred. For this ecosystem and state‐and‐transition models, out of nine vegetation variables considered, the density of immature trees and percentage of exotic understory vegetation cover were the variables most frequently specified as effective to define a threshold or transition. We synthesize findings by presenting a decision tree that provides practical guidance for the development of targeted monitoring strategies for woodland vegetation.